{
  "format": "proofrun-publication",
  "schemaVersion": 1,
  "contentHash": "sha256:ceb35e3680ce91601e1004a220a487ef6beea0fdb36e5448b8ef47e6f5a047e3",
  "preparedAt": "2026-08-08T20:56:05.847Z",
  "editorial": {
    "slug": "terra-s-steadier-physics-beat-sonnet-5-s-audio-edge-in-gravity-desk-87-73",
    "headline": "Terra’s steadier physics beat Sonnet 5’s audio edge in Gravity Desk, 87–73",
    "dek": "In one blind, frozen comparison, contestant B won with high confidence because stronger physics, control, and presentation outweighed A’s more responsive-feeling sound feedback.",
    "status": "draft",
    "publishedAt": null
  },
  "experiment": {
    "protocol": "proofrun-v0.6",
    "runId": "PR-MSKKQ2K6-34",
    "startedAt": "2026-08-08T16:14:07.734Z",
    "identityState": "revealed",
    "evaluationState": "locked",
    "legacyProtocolException": false
  },
  "test": {
    "id": "gravity-desk",
    "title": "Gravity Desk",
    "category": "Physics toy",
    "summary": "A compact canvas toy that exposes technical quality without requiring a full game.",
    "prompt": "Build a playful one-screen physics toy called GRAVITY DESK. Clicking spawns colorful glass-like marbles that fall, bounce off the viewport edges, and collide convincingly enough to feel satisfying. Include Gravity and Clear controls plus a live object count. Use a single offline HTML file with Canvas and vanilla JavaScript, no libraries or assets. Prioritize smoothness, polish, and immediate playability.",
    "origin": "curated",
    "version": 1,
    "hash": "curated:gravity-desk:v1",
    "scope": "quick",
    "runtime": "7-12 min",
    "executionMode": "web-single-file-v2",
    "signals": [
      "Technical correctness",
      "Feel",
      "Performance",
      "Polish"
    ],
    "integrity": "versioned"
  },
  "settings": {
    "reasoningEffort": "high",
    "maxTokens": 32000,
    "temperature": 0.7,
    "htmlStartDeadlineMinutes": 20,
    "writingGraceMinutes": 10
  },
  "verdict": {
    "winner": "B",
    "winnerModel": "openai/gpt-5.6-terra",
    "confidence": "high",
    "rationale": "B's result is overall better because of the stabler physics, more control over the balls and better visuals overall.",
    "scores": {
      "A": 73,
      "B": 87
    }
  },
  "requirements": [
    {
      "id": "requirement-1",
      "label": "Clicking spawns a colorful glass-like marble.",
      "A": "pass",
      "B": "pass"
    },
    {
      "id": "requirement-2",
      "label": "Marbles fall and bounce within the viewport.",
      "A": "pass",
      "B": "pass"
    },
    {
      "id": "requirement-3",
      "label": "Marbles collide convincingly enough for the intended toy.",
      "A": "pass",
      "B": "pass"
    },
    {
      "id": "requirement-4",
      "label": "Gravity and Clear controls work during active simulation.",
      "A": "pass",
      "B": "pass"
    },
    {
      "id": "requirement-5",
      "label": "The live object count remains accurate.",
      "A": "pass",
      "B": "pass"
    },
    {
      "id": "requirement-6",
      "label": "The single-file canvas experience remains smooth, responsive, and offline.",
      "A": "partial",
      "B": "partial"
    }
  ],
  "criteria": [
    {
      "id": "requirement-fit",
      "label": "Requirement fit",
      "description": "How completely and accurately the artifact satisfies the frozen brief.",
      "weight": 20,
      "anchors": {
        "low": "Major requirements missing",
        "mid": "Core brief met with gaps",
        "high": "Complete and exact"
      },
      "A": 10,
      "B": 10
    },
    {
      "id": "physics-feel",
      "label": "Physics & feel",
      "description": "Believability, responsiveness, collision behavior, and the satisfaction of motion.",
      "weight": 25,
      "anchors": {
        "low": "Unconvincing or unstable",
        "mid": "Playful with visible shortcuts",
        "high": "Consistent and satisfying"
      },
      "A": 7,
      "B": 10
    },
    {
      "id": "technical-robustness",
      "label": "Technical robustness",
      "description": "Runtime correctness, edge-case handling, performance, responsiveness, and code reliability.",
      "weight": 25,
      "anchors": {
        "low": "Repeatable failures",
        "mid": "Mostly sound with risks",
        "high": "Stable under scrutiny"
      },
      "A": 6,
      "B": 7
    },
    {
      "id": "visual-design",
      "label": "Visual design",
      "description": "Hierarchy, composition, typography, color, polish, and coherent visual judgment.",
      "weight": 20,
      "anchors": {
        "low": "Confusing or unfinished",
        "mid": "Competent and readable",
        "high": "Distinctive and exceptional"
      },
      "A": 6,
      "B": 7
    },
    {
      "id": "interaction",
      "label": "Interaction & usability",
      "description": "Discoverability, responsiveness, feedback, control quality, and interaction feel.",
      "weight": 10,
      "anchors": {
        "low": "Frustrating or broken",
        "mid": "Usable with rough edges",
        "high": "Immediate and delightful"
      },
      "A": 8,
      "B": 10
    }
  ],
  "contestants": [
    {
      "side": "A",
      "model": "anthropic/claude-sonnet-5",
      "resolvedModel": "anthropic/claude-sonnet-5",
      "provider": "Amazon Bedrock",
      "status": "complete",
      "screenshot": {
        "path": "screenshots/contestant-a-capture-lab-interaction-state-7d2b4f64.png",
        "capturedAt": "2026-08-08T20:28:34.769Z",
        "width": 1278,
        "height": 718,
        "byteLength": 856148,
        "method": "html2canvas-capture-lab-v1"
      },
      "evidenceImages": [
        {
          "id": "EV-691f056f-1bb0-4da5-a34a-5aa97d2b4f64",
          "path": "screenshots/contestant-a-capture-lab-interaction-state-7d2b4f64.png",
          "label": "Capture Lab interaction state",
          "caption": "State manually created by the reviewer in the isolated Capture Lab.",
          "provenance": "harness-current-state",
          "state": "manual",
          "primary": true,
          "capturedAt": "2026-08-08T20:28:34.769Z",
          "width": 1278,
          "height": 718,
          "byteLength": 856148,
          "method": "html2canvas-capture-lab-v1"
        },
        {
          "id": "EV-63ba0080-ef0f-40c4-a95e-c815f264ee85",
          "path": "screenshots/contestant-a-matched-initial-state-f264ee85.png",
          "label": "Matched initial state",
          "caption": "Fresh artifact state captured at the same fixed 1280 × 720 viewport as the other contestant.",
          "provenance": "harness-matched",
          "state": "initial",
          "primary": false,
          "capturedAt": "2026-08-08T20:33:20.267Z",
          "width": 1280,
          "height": 720,
          "byteLength": 772662,
          "method": "html2canvas-isolated-v1"
        }
      ],
      "generation": {
        "startedAt": "2026-08-08T16:14:07.895Z",
        "finishedAt": "2026-08-08T16:15:54.682Z",
        "wallTimeMs": 106787,
        "firstActivityMs": 4157,
        "htmlStartMs": 60381,
        "lastActivityMs": 106755,
        "activityChunks": 161,
        "finishReason": "stop",
        "reasoningEffort": "high"
      },
      "usage": {
        "inputTokens": 293,
        "visibleTokens": 11090,
        "reasoningTokens": 1142,
        "cachedTokens": 0,
        "totalTokens": 12525,
        "cost": 0.122906,
        "costSource": "openrouter-reported"
      },
      "evidence": {
        "artifactIssues": [],
        "runtimeErrors": [],
        "previewAudit": {
          "textCharacters": 83,
          "visibleElements": 14,
          "bodyChildren": 4,
          "documentCharacters": 222092
        },
        "probe": {
          "status": "passed",
          "durationMs": 673,
          "viewport": {
            "width": 536,
            "height": 520
          },
          "actions": [
            {
              "id": "resize",
              "label": "Narrow viewport",
              "status": "passed",
              "detail": "536 x 520 viewport exercised"
            },
            {
              "id": "click",
              "label": "Single click",
              "status": "passed",
              "detail": "Dispatched one complete pointer and mouse click sequence"
            },
            {
              "id": "drag",
              "label": "Short drag",
              "status": "passed",
              "detail": "Dispatched a bounded six-step diagonal drag"
            },
            {
              "id": "burst",
              "label": "Five-point burst",
              "status": "passed",
              "detail": "Dispatched five bounded clicks across the interaction surface"
            },
            {
              "id": "reset",
              "label": "Reset control",
              "status": "passed",
              "detail": "Activated 🧹 Clear"
            }
          ],
          "issues": [],
          "errors": []
        }
      }
    },
    {
      "side": "B",
      "model": "openai/gpt-5.6-terra",
      "resolvedModel": "openai/gpt-5.6-terra",
      "provider": "OpenAI",
      "status": "complete",
      "screenshot": {
        "path": "screenshots/contestant-b-capture-lab-interaction-state-ee19fe65.png",
        "capturedAt": "2026-08-08T20:27:41.587Z",
        "width": 1278,
        "height": 718,
        "byteLength": 1001091,
        "method": "html2canvas-capture-lab-v1"
      },
      "evidenceImages": [
        {
          "id": "EV-a42549c5-8880-4e65-8792-a0b3ee19fe65",
          "path": "screenshots/contestant-b-capture-lab-interaction-state-ee19fe65.png",
          "label": "Capture Lab interaction state",
          "caption": "State manually created by the reviewer in the isolated Capture Lab.",
          "provenance": "harness-current-state",
          "state": "manual",
          "primary": true,
          "capturedAt": "2026-08-08T20:27:41.587Z",
          "width": 1278,
          "height": 718,
          "byteLength": 1001091,
          "method": "html2canvas-capture-lab-v1"
        },
        {
          "id": "EV-c243dacc-84b1-4747-9ec8-1e999226b25a",
          "path": "screenshots/contestant-b-matched-initial-state-9226b25a.png",
          "label": "Matched initial state",
          "caption": "Fresh artifact state captured at the same fixed 1280 × 720 viewport as the other contestant.",
          "provenance": "harness-matched",
          "state": "initial",
          "primary": false,
          "capturedAt": "2026-08-08T20:33:22.262Z",
          "width": 1280,
          "height": 720,
          "byteLength": 87534,
          "method": "html2canvas-isolated-v1"
        }
      ],
      "generation": {
        "startedAt": "2026-08-08T16:14:07.898Z",
        "finishedAt": "2026-08-08T16:15:40.074Z",
        "wallTimeMs": 92176,
        "firstActivityMs": 5216,
        "htmlStartMs": 51017,
        "lastActivityMs": 92126,
        "activityChunks": 8668,
        "finishReason": "stop",
        "reasoningEffort": "high"
      },
      "usage": {
        "inputTokens": 176,
        "visibleTokens": 7537,
        "reasoningTokens": 3777,
        "cachedTokens": 0,
        "totalTokens": 11490,
        "cost": 0.13612,
        "costSource": "upstream-estimate"
      },
      "evidence": {
        "artifactIssues": [],
        "runtimeErrors": [],
        "previewAudit": {
          "textCharacters": 153,
          "visibleElements": 24,
          "bodyChildren": 5,
          "documentCharacters": 230050
        },
        "probe": null
      }
    }
  ],
  "disclosure": {
    "rawModelResponsesIncluded": false,
    "contestantSourceIncluded": false,
    "privateReviewerNotesIncluded": false,
    "credentialsIncluded": false,
    "generatedHtmlExecutedByPublicRecord": false,
    "editorialAssistance": {
      "evidenceAnalysis": {
        "model": "openai/gpt-5.6-sol",
        "promptVersion": "evidence-analyst-v2",
        "blind": true,
        "cost": 0.31811875,
        "acceptedFindings": 6,
        "rejectedFindings": 3
      },
      "articleDrafting": {
        "model": "openai/gpt-5.6-sol",
        "promptVersion": "article-copilot-v1",
        "cost": 0.13572875,
        "humanRevised": false,
        "humanApproved": true
      }
    }
  }
}